Trajectories of seasonal influenza vaccine uptake among French people with diabetes: a nationwide retrospective cohort study, 2006-2015
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Bocquier et al. BMC Public Health (2019) 19:918
https://doi.org/10.1186/s12889-019-7209-z
RESEARCH ARTICLE Open Access
Trajectories of seasonal influenza vaccine
uptake among French people with
diabetes: a nationwide retrospective cohort
study, 2006–2015
Aurélie Bocquier1,2,3*, Sébastien Cortaredona1,2, Lisa Fressard1,2,3, Pierre Loulergue4,5,6,7, Jocelyn Raude8,9,
Ariane Sultan10,11, Florence Galtier4,12 and Pierre Verger1,2,3,4
Abstract
Background: Annual seasonal influenza vaccination (SIV) is recommended for people with diabetes, but their SIV
rates remain far below public health targets. We aimed to identify temporal trajectories of SIV uptake over a 10-year
period among French people with diabetes and describe their clinical characteristics.
Methods: We identified patients with diabetes in 2006 among a permanent, representative sample of beneficiaries
of the French National Health Insurance Fund. We followed them up over 10 seasons (2005/06–2015/16), using SIV
reimbursement claims and group-based trajectory modelling to identify SIV trajectories and to study sociodemographic,
clinical, and healthcare utilization characteristics associated with the trajectories.
Results: We identified six trajectories. Of the 15,766 patients included in the model, 4344 (28%) belonged to the
“continuously vaccinated” trajectory and 4728 (30%) to the “never vaccinated” one. Two other trajectories showed a
“progressive decrease” (2832, 18%) or sharp “postpandemic decrease” (1627, 10%) in uptake. The last two trajectories
(totalling 2235 patients, 14%) showed an early or delayed “increase” in uptake. Compared to “continuously vaccinated”
patients, those in the “progressively decreasing” trajectory were older and those in all other trajectories were younger with
fewer comorbidities at inclusion. Worsening diabetes and comorbidities during follow-up were associated with the
“increasing” trajectories.
Conclusions: Most patients with diabetes had been continuously vaccinated or never vaccinated and thus had stable SIV
behaviours. Others adopted or abandoned SIV. These behaviour shifts might be due to increasing age, health events, or
contextual factors (e.g., controversies about vaccine safety or efficacy). Healthcare professionals and stakeholders should
develop tailored strategies that take each group’s specificities into account.
Keywords: Diabetes mellitus, Influenza vaccines, Cohort studies, Administrative claims, Healthcare
Background nonetheless below WHO’s target of 75% in most Western
Because people with diabetes are at increased risk of severe countries [5–7], especially in France (26% in 2015/16
complications linked to seasonal influenza [1], the World among those < 65 years) [8].
Health Organization (WHO) and many national guidelines Although SIV must be repeated annually, few cohort
[2–4] recommend they receive annual seasonal influenza studies have explored the course of SIV behaviours for sev-
vaccination (SIV). The SIV rate in this population is eral consecutive years. They have found evidence for both
stable SIV behaviours and behaviour shifts (e.g., stopping
SIV) [9, 10], suggesting that distinct temporal patterns (tra-
* Correspondence: aurelie.bocquier@inserm.fr
1
Aix Marseille Univ, IRD, AP-HM, SSA, VITROME, 19-21 Boulevard Jean Moulin, jectories) of SIV behaviour may exist. Trajectories have
13385 Marseille Cedex 05, France been studied for other significant aspects of diabetes man-
2
IHU-Méditerranée Infection, Marseille, France agement (e.g., glycemic control and adherence to oral
Full list of author information is available at the end of the article
© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Bocquier et al. BMC Public Health (2019) 19:918 Page 2 of 8
hypoglycemic agents [11, 12]), but there is no literature Seasonal influenza vaccine uptake
about SIV trajectories. Identifying trajectories of SIV behav- For each individual and each SIV season n/n + 1 (temporal
iour among people with diabetes may help identify patients statistical units in our analyses), we constructed a binary
on whom prevention efforts should concentrate, with variable “SIV uptake” (yes, no), based on SIV deliveries
tailored communication and behaviour-change strategies. (Additional file 2: Table S2) recorded between September 1
Based on reimbursement data, this article sought to: of year n and March 31 of year n + 1. Each SIV delivered in
1) identify temporal patterns (trajectories) of SIV uptake a community pharmacy is recorded in the NHIF database.
among French people with diabetes over 10 consecutive in-
fluenza seasons (2005/06 to 2015/16) and determine their Characteristics of the study population
prevalence; and 2) study the sociodemographic, clinical, and To describe the diabetes type and treatments, we con-
healthcare utilization characteristics associated with them. structed for each year of follow-up a 5-category variable
based on LTI status and reimbursement claims for antidia-
betic drugs recorded during the 6 months before the start
Methods of season n/n + 1. Using these annual variables, we built a
Study design and data source variable of “diabetes treatment intensification” (yes/no)
We conducted a retrospective cohort study in the Perman- during follow-up. “Intensification” was defined by at least
ent Sample of Beneficiaries (Echantillon Généraliste des Bén- one of the following modifications: from no antidiabetic
éficiaires, EGB). The EGB, set up in 2005, is a permanent, drug to at least one antidiabetic drug; from only one to at
representative, and open national random sample of 1/97th least two noninsulin antidiabetic drugs or insulin; from at
of persons currently affiliated with one of the three major na- least two noninsulin antidiabetic drugs to insulin.
tional health insurance funds in France [13]. At the time of To measure comorbidities, we calculated for each year of
extraction (August 2017), it included 804,089 beneficiaries. follow-up an individual chronic condition score (ICC) based
For this study, we extracted data for salaried workers on drug deliveries according to a previously published
(including those who are retired) only (about 86% of methodology [14]. Then we built a 3-category variable
the French population [13], covered by the French describing the course of the ICC score from the first to the
National Health Insurance Fund, NHIF) because last year of follow-up (decreasing, increasing, or stable) and
people affiliated with the other insurance funds were included it in our analysis, as a time-stable variable [15].
only included in the EGB in 2011. Each cohort member’s number of hospital stays for
Data include age, gender, district of residence, reimburse- each year for diabetes, diabetes complications, influenza,
ment claims for consultations with private healthcare pro- and influenza complications was extracted, as were the
fessionals, medical procedures (e.g., laboratory tests), drugs numbers of visits – separately – with general practi-
purchased in the community (classified by Anatomical tioners (GPs), endocrinologists, and cardiologists. GPs
Therapeutic Chemical (ATC) codes), and long-term illness are responsible for the management of most patients
(LTI) status, recorded by expert physicians according to the with type 2 diabetes [9] and for referral to specialists.
International Classification of Diseases (ICD-10). LTI status We also extracted changes of GP during follow-up.
is granted to beneficiaries with long-term and costly The NHIF sends free vaccination vouchers each season to
diseases and exempts them, regardless of income level, individuals aged 65 years or older and to those patients with
from copayments for related medical care [14]. Since 2006, diabetes with an LTI status: we constructed a 3-category
data regarding diagnoses associated with admissions to variable to describe receipt of this voucher (Additional file 3:
French public or private hospitals are also available. Table S3). The voucher enables these patients to obtain the
The NHIF granted us authorization to access the EGB, vaccine free of charge at the pharmacy, without a doctor’s
in accordance with French law. prescription. They must then make an appointment with ei-
ther a doctor or a nurse for its administration.
Study population Statistical analysis
Using our adaptation of an NHIF algorithm [8] based on We ran group-based trajectory (GBT) modelling to iden-
LTI status, hospitalization diagnoses, and reimbursement tify subgroups of individuals with similar patterns of SIV
claims for antidiabetic drugs or hemoglobin A1c (HbA1c) dispensing over time during the 10-year follow-up period.
assays (Additional file 1: Table S1), we selected all individ- GBT modelling is a data-driven semiparametric method
uals residing in mainland France identified as treated for designed to analyze the evolution of an outcome over time
diabetes in 2006. We followed them up over 10 seasonal and to identify, within a population with unobserved het-
vaccination campaigns. Those who died or withdrew from erogeneity, distinct clusters of individuals following similar
the NHIF during the follow-up period were censored at the trajectories of behaviors related to this outcome [16, 17].
start of the year of the event. It makes it possible to select the model with an optimalBocquier et al. BMC Public Health (2019) 19:918 Page 3 of 8
number of distinct trajectories that most appropriately In trajectory 1 (“continuously vaccinated”, 28% of the
represent the heterogeneity in the population [15]. To cohort), the SIV-uptake rate started at 92% at inclusion,
compare the models’ goodness of fit, we used the Bayesian then exceeded 97% throughout follow-up, with numbers
information criterion (BIC) and individual posterior class- of SIV injections over the 10-season follow-up ranging
membership probabilities (i.e., the probability of belonging between 8 and 10. In trajectory 2 (“progressively less
to a trajectory given the information collected). Starting vaccinated”, 18%), SIV uptake exceeded 95% at inclusion,
with a one-trajectory solution, we added one trajectory at finally dropping to 59% in 2015/16 (range of SIV injec-
a time, testing each model fit and balancing it with our ob- tions: [6–9]). The uptake in trajectory 3 (“postpandemic
jective of identifying distinct and interpretable trajectories. decreasingly vaccinated”, 10%) was relatively high (63–
The prevalence of each trajectory and the relevance of the 73%) and stable until the 2009/10 influenza A(H1N1) pan-
solutions were also considered, as recommended by Nagin demic season; it dropped by 33 percentage points in 2010/
and Odgers [18]. To determine the order of polynomials 11 (range of SIV injections: [2–8]). Trajectory 4 (“early in-
for all trajectories, we started with third-degree polyno- creasingly vaccinated”, 9%) began with a very low SIV-
mials and used the standard operating procedure, i.e. step- uptake rate at inclusion and then immediately and rapidly
wise elimination of non-significant polynomial higher increased, stabilizing around 90% in 2011/12 (range of
orders [19]. Early applications of GBT modelling have as- SIV injections: [4–9]). Trajectory 5 (“late increasingly vac-
sumed that all attrition (including both loss to follow-up cinated”, 5%) looked like trajectory 4 with the increasing
and mortality) is randomly distributed among all trajector- phase shifted forward several years (range of SIV injec-
ies. A recent enhancement of the GBT approach enables tions: [2–6]). The individuals with trajectory 6 (“never vac-
the joint modelling of the outcome of interest and non- cinated”, 30%) had very low SIV-uptake rates throughout
random missingness [20, 21]. Using this methodology, we follow-up (range of SIV injections: [0–2]).
were able to model attrition probabilities (mortality repre-
sented the vast majority of attrition in our study) jointly
with the estimation of SIV trajectories. Risk factors for SIV-uptake trajectory memberships
The demographic, clinical, and healthcare utilization fac- With the “continuously vaccinated” trajectory as the refer-
tors were added to the model as predictors of trajectory ence (Table 2), the probability of belonging to the “progres-
group membership. This joint estimation of trajectories and sively less vaccinated” trajectory was higher for individuals
predictors of the probability of group membership allowed aged 65 years or older at inclusion, those receiving no anti-
us to take into account the uncertainty in participants’ trajec- diabetic drug, with high comorbidity scores at inclusion
tory group membership [15]. We used Zhang’s correction to and remaining stable during follow-up, hospitalized for
estimate adjusted risk ratios from the estimated ratios [22]. influenza during follow-up, and seeing GPs frequently. It
All statistical analyses were performed with SAS statistical was lower among women, for those with intensified
software, version 9.4 (SAS Institute Inc., Cary, NC). GBT diabetes treatment, seeing endocrinologists frequently, and
analyses were conducted with the TRAJ procedure [16]. changing GPs during follow-up.
The remaining four trajectories (“postpandemic decreas-
Results ingly vaccinated”, “early”/“late increasingly vaccinated”,
Study population characteristics (Table 1) and “never vaccinated”) shared several characteristics. The
Of the 17,259 subjects with diabetes included in the co- probability of belonging to these four trajectories was
hort in 2006, 46% were women; mean age at inclusion was higher in patients receiving no antidiabetic drug at inclu-
65.0 ± 13.7 years. About 10% were identified with type 1 sion and lower in those aged 65 years or older, with more
diabetes at inclusion; only 70% had LTI status for diabetes comorbidities at inclusion, and with frequent visits with
then. Over the 10-year follow-up, 31% of the initial cohort specialists during follow-up. These trajectories also
died (Additional file 4: Table S4), for a death rate of 36‰ showed some specificities. The probability of belonging to
person-years, and 3% were lost to follow-up. the “postpandemic decreasingly vaccinated” trajectory was
higher for women and individuals hospitalized for diabetes
SIV-uptake trajectories (Fig. 1) or influenza; it was lower for those with worsening comor-
Based on the BIC values, the fit of the models improved as bidities. The probability of belonging to the “early” or “late
the number of trajectories modeled increased. From a increasingly vaccinated” trajectories was higher for those
seven-trajectory solution and after, the prevalence of some with worsening diabetes and comorbidities during follow-
trajectories was very low and results were difficult to inter- up, and those hospitalized for influenza (for the “early”
pret. Accordingly, the solution that offered the best com- trajectory only); it was lower for individuals with type 1
promise between parsimony, fit, and interpretability was a diabetes. Finally, the probability of belonging to the “never
six-trajectory solution. Classification quality was good for all vaccinated” trajectory was higher for women and for indi-
six (mean posterior class-membership probability > 0.82). viduals with stable comorbidities, and lower for those withBocquier et al. BMC Public Health (2019) 19:918 Page 4 of 8
Table 1 Study cohort characteristics during the first and last seasons n/n + 1 of follow-up (EGB, France, 2006/07–2015/16)
2006/07 (n = 17,259) 2015/16 (n = 11,440)a
b
% %b
Sociodemographic characteristics
Age (years) on 12.31.n – mean (SD) 65.0 (13.7) 70.5 (12.8)
Women 46.1 47.1
Clinical characteristics
Type and treatment of diabetes
Type 1 diabetesc 9.2 11.2
Other types -- no antidiabetic drug 15.8 9.2
Other types - only one noninsulin 35.3 22.0
antidiabetic drug
Other types -- ≥ 2 noninsulin antidiabetic 28.2 33.6
drugs
Other types -- insulin treatment ± antidiabetic 11.5 24.0
drugs
Weighted individual chronic condition scored – mean (SD) 0.8 (0.5) 0.9 (0.5)
Annual rate of hospitalization for diabetes or its complicationse 6.2 4.3
e
Annual rate of hospitalization for influenza or its complications 0.7 1.4
Healthcare utilization
Annual number of consultationsf with -- mean (SD)
General practitioner 8.6 (7.3) 7.9 (6.4)
Endocrinologist 0.3 (1.2) 0.4 (1.2)
Cardiologist 0.5 (1.5) 0.5 (1.4)
Change of general practitioner 4.0 10.3
Received free vaccination voucher for diabetesg 69.9 90.3
SD standard deviation
a
Among all patients included in the cohort, 5266 (30.5%) died and 553 (3.2%) were lost during follow-up. Mean follow-up time: 8.24 ± 2.90 seasons
b
Otherwise stated
c
People with type 1 diabetes were those with long-term illness status for type 1 diabetes (E10 according to the ICD-10) and treated by insulin at inclusion
d
The individual chronic condition score (ICC) was calculated as a weighted sum of 21 chronic conditions. Weights account for the severity of each condition in the
score calculation (ICC range in study cohort: min = 0; max = 3.7)
e
At least 1 hospitalization between 09.01.n-1 and 08.31.n
f
Number of consultations between 09.01.n-1 and 08.31.n
g
To identify people with diabetes, the National Health Insurance Fund uses only their Long-Term Illness (LTI) status on September 1 of each year. Nonetheless, not
all patients with diabetes (especially those with diabetes other than type 1) receive the voucher, because some who should have LTI status do not apply for it
type 1 diabetes, with worsening comorbidities, frequent with fewer comorbidities at inclusion. The “increasing”
healthcare utilization, and changing GPs during follow-up. trajectories were positively associated with the worsen-
ing of diabetes and comorbidities during follow-up.
Discussion
Key findings
Overall, this study shows remarkable inertia in behav- Strengths and limitations
ioural patterns, with 28% of the subjects continuously The strengths of this study include its 10-year follow-up,
vaccinated and 30% never vaccinated from 2006/07 to the longest for any study examining SIV behaviours over
2015/16. For two other trajectories, the SIV-uptake rate time [6, 9, 10], and its large sample size. Moreover, our al-
decreased during follow-up, either progressively (18%) gorithm to identify patients with diabetes was more sensi-
or more sharply after the 2009/10 season (10%), while tive and allowed earlier identification than an algorithm
the SIV-uptake rate rose for the last two trajectories (ac- based solely on LTI [8]. We used vaccine deliveries, which
counting for only 14% of patients). Compared to “con- are more reliable than self-reported vaccination behaviour
tinuously vaccinated” people, those in the “progressively [23]. The dropout extension of the group-based trajectory
decreasing” trajectory were older; those in the “postpan- modelling allowed us to control for potential selection
demic decreasing”, “increasing”, and “never” vaccinated biases due to non-random participant attrition (especially
trajectories were younger than the reference category those due to mortality, Additional file 3: Table S3) [20].Bocquier et al. BMC Public Health (2019) 19:918 Page 5 of 8
Fig. 1 Observed (solid lines) and predicted (dashed lines) probability of seasonal influenza vaccine uptake among people with diabetes during
each season of follow-up for each of six classes identified by the group-based trajectory modela (EGB, France, 2006/07–2015/16, n = 15,766b).
a
Third-degree polynomials were used for the specifications of all trajectories, except for the “Early increasingly vaccinated” trajectory, for which a
second-degree polynomial was used. b Among individuals with at least two full years of follow-up (n = 15,766, 90.2%), to enable calculation of
two variables included in the model (i.e., diabetes treatment intensification and course of weighted individual chronic condition score
during follow-up)
We acknowledge some limitations. Vaccinations that follow-ups [9]. When health protective behaviours must
took place during occupational medicine visits or at vac- be regularly repeated in stable contexts, patients’ re-
cination centres or some nursing homes that buy vaccines sponses to their healthcare workers’ recommendations
for their residents (fewer than 20% of all nursing homes may be performed almost automatically, without either
[24]) are not recorded in the French NHIF databases. conscious decision-making or thinking [29]. This inter-
However, these limitations are unlikely to affect our re- pretation is in line with recent advances in behavioural
sults substantially as the vast majority of vaccinations in sciences showing that “much human behaviour is auto-
France are administered by private healthcare workers matic, cued by environmental stimuli” [30]. We may as-
and are thus recorded in these databases [25]. As SIV sume that subjects continuously vaccinated were aware
behaviour varies by socioeconomic characteristics [26], of their vulnerability to influenza (due to age and/or co-
our results cannot be extrapolated to population categor- morbidities [9]) before our follow-up began. Another hy-
ies not covered by the NHIF (e.g., farmers, the self- pothesis is that receiving a free voucher each year at
employed) or the very few people without insurance; least as early as inclusion (Additional file 3: Table S3)
nonetheless, the NHIF covers 86% of the French popula- and regular medical consultations may foster SIV behav-
tion. Several socioeconomic (e.g., educational level) and iours because they act as reminders [31] and the
clinical (e.g., diabetes complications) characteristics are vouchers may facilitate access to the vaccine [8]. Con-
not recorded in NHIF databases and therefore could not versely, studies show that continuously refusing SIV is
be studied. Specifically, no data about individuals’ know- often associated with attitudes of risk neutralization
ledge, attitudes or perceptions towards SIV (e.g., beliefs (e.g., comparing SI with other infectious diseases, feeling
about SIV efficacy, side effects) were available, although “mentally and physically” able to resist SI) [28]. Oppor-
they are important drivers of SIV behaviours [27, 28] and tunities might also have been missed: we estimated that,
thus probably differ according to trajectories. at inclusion, 30% of patients with diabetes did not re-
ceive free vouchers because they did not benefit from
Interpretation of the findings LTI status. These patients can obtain a voucher from
Our finding that most people with diabetes had stable their doctor but this makes their pathway to vaccination
SIV behaviours is consistent with results from previous still more complex as it requires first a doctor’s appoint-
qualitative [28] and quantitative studies with shorter ment to get a free vaccine voucher, then a trip to theBocquier et al. BMC Public Health (2019) 19:918 Page 6 of 8
Table 2 Risk factors for membership in SIV-uptake trajectories – group-based trajectory model, multinomial logistic regressiona (EGB,
France, 2006/07–2015/16, n = 15,766b)
Trajectory (ref. 1. Continuously vaccinated - n = 4344 (27.6%)
2. Progressively 3. Post pandemic 4. Early increasingly 5. Late increasingly 6. Never
less vaccinated decreasingly vaccinated vaccinated vaccinated vaccinated
n = 2832 n = 1627 n = 1472 n = 763 n = 4728
18.0% 10.3% 9.3% 4.8% 30.0%
Adjusted risk ratio [95% confidence interval]
Sociodemographic characteristics
Age at inclusion > 65 2.89 [2.47;3.33] 0.68 [0.60;0.78] 0.33 [0.28;0.38] 0.14 [0.11;0.18] 0.56 [0.52;0.61]
Women 0.82 [0.75;0.91] 1.38 [1.26;1.51] 0.99 [0.88;1.10] 0.98 [0.83;1.14] 1.09 [1.05;1.14]
Clinical characteristics
Type and treatment of diabetes at inclusion
Type 1c 0.99 [0.83;1.16] 1.11 [0.95;1.30] 0.76 [0.59;0.97] 0.53 [0.37;0.76] 0.85 [0.76;0.94]
Other types -- no antidiabetic drug 1.18 [1.01;1.36] 1.61 [1.39;1.85] 3.04 [2.77;3.29] 2.10 [1.69;2.56] 1.62 [1.54;1.68]
d
Diabetes treatment intensification during 0.76 [0.68;0.84] 1.04 [0.94;1.15] 1.23 [1.09;1.39] 1.25 [1.06;1.47] 0.96 [0.91;1.01]
follow-up
Weighted individual chronic condition 1.16 [1.04;1.28] 0.87 [0.78;0.97] 0.77 [0.68;0.87] 0.81 [0.68;0.96] 0.75 [0.71;0.80]
scoree at inclusion ≥ median
Course of weighted individual chronic
condition scoree during follow-up
Stable 1.33 [1.06;1.60] 0.81 [0.55;1.13] 0.68 [0.40;1.11] 0.83 [0.40;1.62] 1.45 [1.33;1.57]
Increasing 0.99 [0.90;1.08] 0.73 [0.66;0.81] 1.19 [1.05;1.34] 1.21 [1.02;1.43] 0.89 [0.83;0.94]
Hospitalized during follow-up
For diabetes and its complications 1.05 [0.94;1.16] 1.23 [1.11;1.37] 1.01 [0.88;1.15] 1.02 [0.85;1.22] 1.05 [1.00;1.11]
For influenza and its complications 1.32 [1.16;1.48] 1.44 [1.22;1.67] 1.46 [1.20;1.75] 1.30 [0.91;1.81] 0.98 [0.87;1.09]
Healthcare utilization
Frequent consultations during follow-up,
with:
General practitioner 1.92 [1.76;2.07] 1.02 [0.91;1.14] 0.95 [0.83;1.07] 0.80 [0.66;0.96] 0.85 [0.80;0.90]
Endocrinologist 0.77 [0.65;0.91] 0.70 [0.59;0.83] 0.82 [0.69;0.97] 0.64 [0.49;0.83] 0.84 [0.78;0.91]
Cardiologist 1.00 [0.90;1.11] 0.79 [0.69;0.92] 0.87 [0.74;1.01] 0.71 [0.55;0.91] 0.79 [0.73;0.86]
Change of general practitioner during 0.73 [0.66;0.81] 0.98 [0.89;1.08] 0.94 [0.84;1.05] 1.13 [0.97;1.31] 0.93 [0.88;0.97]
follow-up
Reference groups. Age: “≤ 65 years”; gender: “men”; type and treatment of diabetes at inclusion: “other types -- ≥ 1 antidiabetic drug”; diabetes treatment
intensification: “no”; weighted individual chronic condition score at inclusion: “< median”; course of weighted individual chronic condition score: “decreasing”;
hospitalized during follow-up: “no”; consultations during follow-up: “number of consultations < median”; change of general practitioner: “no”
Boldface indicates statistical significance (p ≤ 0.05)
a
Model adjusted for all variables displayed in the Table, as well as for district of residence (results not displayed): Paris region, northwest, northeast, southeast, and
southwest. The variable “Received the free vaccination voucher” was not included in the model due to strong correlation with age (all people aged 65 years or
older receive this voucher)
b
Among individuals with at least two full years of follow-up (n = 15,766, 90.2%), to enable calculation of two variables included in the model (i.e., diabetes
treatment intensification and course of weighted individual chronic condition score during follow-up).
c
People with type 1 diabetes were those with long-term illness status for type 1 diabetes (E10 according to the ICD-10) and treated by insulin at inclusion
d
“Intensification” was defined by at least one of the following modifications during follow-up: from no antidiabetic drug to at least one antidiabetic drug; from
only one to at least two noninsulin antidiabetic drugs or insulin; from at least two noninsulin antidiabetic drugs to insulin
e
The individual chronic condition score (ICC) was calculated as a weighted sum of 21 chronic conditions. Weights account for the severity of each condition in the
score calculation (ICC range in study cohort: min = 0; max = 3.7)
pharmacy to pick up the vaccine, and then a second ap- cohort) may imply a progressive phasing-out of SIV among
pointment for the actual injection. the frail elderly. This may result from doubts among pa-
Nonetheless, the shifts in SIV behaviour among distinct tients, their relatives and/or their doctors about the benefits
groups of patients suggest different underlying mechanisms of SIV in the oldest populations, due to the scientific debate
of behaviour change. The characteristics of individuals in the and its media coverage regarding SIV effectiveness and
“progressively less vaccinated” trajectory (the oldest in our immunosenescence [32, 33]. Our results that patients in theBocquier et al. BMC Public Health (2019) 19:918 Page 7 of 8
“progressively less vaccinated” trajectory had less frequent Additional files
consultations with endocrinologists and antidiabetic treat-
ment might also suggest that diabetes itself and prevention Additional file 1: Table S1. Algorithm used to identify individuals with
diabetes in the study. (DOCX 46 kb)
of its complications has become a lower priority among
Additional file 2: Table S2. List of seasonal influenza vaccines selected
these patients. for the study. (DOCX 45 kb)
The “postpandemic decreasingly vaccinated” trajectory Additional file 3: Table S3. Prevalence and characteristics of trajectories
strongly echoes the fall in SIV coverage observed in most identified by the six-class group-based trajectory model. (DOCX 49 kb)
at-risk groups in France after the 2009 A(H1N1) pan- Additional file 4: Table S4. Characteristics of cohort members who
demic season [8]. This drop has been linked to the con- died during the follow-up period. (DOCX 48 kb)
troversies about the safety and effectiveness of the
A(H1N1) vaccine surrounding the French mass vaccin- Abbreviations
BIC: Bayesian information criterion; EGB: Echantillon Généraliste des
ation campaign against the pandemic [8]. The overrep- Bénéficiaires, Permanent Sample of Beneficiaries; GBT: group-based trajectory;
resentation of women in this trajectory is consistent GPs: general practitioners; ICC: individual chronic condition score; LTI: long-term
with gender differences in vaccine hesitancy found for illness; NHIF: National Health Insurance Fund; SIV: seasonal influenza vaccination
other vaccines [26]. Acknowledgments
Finally, our results regarding the “early/late increas- We thank Florence Garry (National Health Insurance Fund, Cnam-TS) and Professor
ingly vaccinated” trajectories suggest that adverse health Jean-Luc Pasquié (CHU Montpellier) for their help in defining algorithms to identify
the study population, Daniel Levy-Bruhl (French Public Health Agency) for his contri-
events (e.g., intensification of diabetes treatment, wors- bution to the study design, and Jo Ann Cahn for her help in editing the
ening comorbidities) may foster or trigger adoption of manuscript.
SIV, which is in line with previous findings [9]. Finally,
the percentages of subjects receiving free vouchers for Authors’ contributions
AB, SC, LF, FG, and PV contributed to the study design and the interpretation
the first time during follow-up rather than at baseline of data. LF and SC conducted data analysis. PL, JR, and AS contributed to the
were highest in these trajectories (Additional file 3: Table interpretation of the data. AB and SC wrote the first draft of the manuscript.
S3). This finding suggests that offering a voucher might All authors contributed to further versions of the manuscript. All authors
have read and approved the manuscript.
foster positive behaviour change [8, 31].
Funding
This study was conducted with the financial support of the Institut de Recherche en
Santé Publique (IReSP) as part of its 2014 general call for research projects
Conclusions (convention no. AAP-2015-03). This study was also supported by the Institut
Our results support the need for a change of the prevention Hospitalo-Universitaire (IHU) Méditerranée Infection, the National Research Agency
paradigm from undifferentiated interventions to interven- under the program « Investissements d’avenir », reference ANR-10-IAHU-03, the Ré-
gion Provence Alpes Côte d’Azur and European funding FEDER PRIMI.
tions that take the specificities of each trajectory into ac-
count. Evidence that SIV strongly decreases among frail Availability of data and materials
elderly with diabetes reminds us of the importance of im- The data that support the findings of this study are available from the National
Health Insurance Fund but restrictions apply to the availability of these data,
proving healthcare professionals’ perceptions of the benefit- which were used under license for the current study and so are not publicly
risk balance of SIV. Practice guidelines could provide add- available. Data are however available from the authors upon reasonable request
itional facts about SIV of the elderly, recognizing issues of and with permission of the National Health Insurance Fund.
immunosenescence and lower SIV efficacy at the individual Ethics approval and consent to participate
level, but emphasizing its importance at the community The study was performed with reimbursement data from the National System
level. Increasing the participation of patients’ relatives in of Health Data (Système National des Données de Santé), in accordance with
the General Conditions of Use of the Portal and the Data (Conditions Générales
patient education for chronic conditions might also be d’Utilisation du Portail et des Données) (version 3.0). Because the study was
effective in enhancing the SIV uptake of both relatives and performed in accordance with the Article L1461–1 (paragraph III.6) of the
the elderly (i.e., indirect and direct protection) [34]. Other French Public Health Code (Code de la santé publique) for public health
purposes with fully anonymized data, there were no further requirements for
countries have chosen to vaccinate children –an important ethical approval, consent to participate or data protection agency approval.
SI virus reservoir—however [3]. Our study also suggests
that health events may represent critical periods during Consent for publication
which healthcare workers might successfully address vac- Not applicable.
cine hesitancy; they should be more aware of these oppor- Competing interests
tunities during patient care. Further interventional research The authors declare that they have no competing interests.
is needed to design more effective interventions to tackle
Author details
vaccine hesitancy regarding SIV. In particular, the use of 1
Aix Marseille Univ, IRD, AP-HM, SSA, VITROME, 19-21 Boulevard Jean Moulin,
tailored communication styles (e.g., presumptive or open 13385 Marseille Cedex 05, France. 2IHU-Méditerranée Infection, Marseille,
approaches and motivational interviewing) that consider France. 3ORS PACA Observatoire Régional de la Santé Provence-Alpes-Côte
d’Azur, Marseille, France. 4INSERM, F-CRIN Innovative Clinical research
patients’ characteristics (e.g., vaccine hesitancy and educa- Network in vaccinology (I-Reivac), GH Cochin Broca Hôtel Dieu, 75014 Paris,
tional level) deserve more research [35]. France. 5Université Paris Descartes, Sorbonne Paris cité, Paris, France. 6InsermBocquier et al. BMC Public Health (2019) 19:918 Page 8 of 8
CIC 1417, Paris, France. 7Assistance Publique Hôpitaux de Paris, CIC 17. Verger P, Mmadi Mrenda B, Cortaredona S, Tournier M, Verdoux H.
Cochin-Pasteur, Paris, France. 8EHESP Rennes, Université Sorbonne Paris Cité, Trajectory analysis of anxiolytic dispensing over 10 years among new users
Paris, France. 9Unité des Virus Emergents (UVE: Aix-Marseille Univ – IRD 190 – aged 50 and older. Acta Psychiatr Scand. 2018;137:328–41. https://doi.org/
Inserm 1207 – IHU Méditerranée Infection), Marseille, France. 10.1111/acps.12858.
10
Endocrinology-Diabetology-Nutrition Department, University Hospital, 18. Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research.
Montpellier, France. 11PhyMedExp, University of Montpellier CNRS INSERM, Annu Rev Clin Psychol. 2010;6:109–38. https://doi.org/10.1146/annurev.
Montpellier, France. 12CIC 1411 CHU Montpellier Hôpital Saint Eloi, clinpsy.121208.131413.
Montpellier, France. 19. Jones BL [Internet]. traj: group-based modeling of longitudinal data. 2018
[cited 2019 May 14]. Available from: https://www.andrew.cmu.edu/user/
Received: 22 November 2018 Accepted: 20 June 2019 bjones/strtxmpl1.htm.
20. Haviland AM, Jones BL, Nagin DS. Group-based trajectory modeling
extended to account for nonrandom participant attrition. Sociol Methods
Res. 2011;40:367–90. https://doi.org/10.1177/0049124111400041.
References 21. Martin LG, Zimmer Z, Lee J. Foundations of activity of daily living
1. Mertz D, Kim TH, Johnstone J, Lam P-P, Science M, Kuster SP, et al. trajectories of older Americans. J Gerontol Ser B. 2017;72:129–39. https://doi.
Populations at risk for severe or complicated influenza illness: systematic org/10.1093/geronb/gbv074.
review and meta-analysis. BMJ. 2013;347:f5061. 22. Zhang J, Yu KF. What’s the relative risk?: a method of correcting the odds
2. World Health Organization [Internet]. Fact sheet on influenza (seasonal) ratio in cohort studies of common outcomes. JAMA. 1998;280:1690–1.
2018 [cited 2019 may 14]. Available from: https://www.who.int/en/news- https://doi.org/10.1001/jama.280.19.1690.
room/fact-sheets/detail/influenza-(seasonal). 23. Jiménez-García R, Hernandez-Barrera V, Rodríguez-Rieiro C, Carrasco Garrido
3. Rizzo C, Rezza G, Ricciardi W. Strategies in recommending influenza P, López de Andres A, Jimenez-Trujillo I, et al. Comparison of self-report
vaccination in Europe and US. Hum Vaccines Immunother. 2018;14:693–8. influenza vaccination coverage with data from a population based
https://doi.org/10.1080/21645515.2017.1367463. computerized vaccination registry and factors associated with discordance.
4. American Diabetes Association. 3. Comprehensive medical evaluation and Vaccine. 2014;32:4386–92. https://doi.org/10.1016/j.vaccine.2014.06.074.
assessment of comorbidities: standards of medical Care in Diabetes—2018. 24. Agence technique de l’information sur l’hospitalisation. Les coûts en
Diabetes Care. 2018;41:S28–37. https://doi.org/10.2337/dc18-S003. établissement d’hébergement pour personnes âgées dépendantes. Lyon:
5. Loerbroks A, Stock C, Bosch JA, Litaker DG, Apfelbacher CJ. Influenza ATIH; 2013.
vaccination coverage among high-risk groups in 11 European countries. Eur 25. Institut de veille sanitaire. Mesure de la couverture vaccinale en France.
J Pub Health. 2012;22:562–8. https://doi.org/10.1093/eurpub/ckr094. Bilan des outils et méthodes en l’an 2000. Saint-Maurice: Institut de veille
6. Jiménez-Garcia R, Lopez-de-Andres A, Hernandez-Barrera V, Gómez- sanitaire; 2001.
Campelo P, San Andrés-Rebollo FJ, de Burgos-Lunar C, et al. Influenza 26. Rey D, Fressard L, Cortaredona S, Bocquier A, Gautier A, Peretti-Watel P, et
vaccination in people with type 2 diabetes, coverage, predictors of uptake, al. Vaccine hesitancy in the French population in 2016, and its association
and perceptions. Result of the MADIABETES cohort a 7 years follow up with vaccine uptake and perceived vaccine risk–benefit balance. Euro
study. Vaccine. 2017;35:101–8. https://doi.org/10.1016/j.vaccine.2016.11.039. Surveill. 2018;23. https://doi.org/10.2807/1560-7917.ES.2018.23.17.17-00816.
7. Villarroel MA, Vahratian A. Vaccination coverage among adults with 27. Nagata JM, Hernández-Ramos I, Kurup AS, Albrecht D, Vivas-Torrealba C,
diagnosed diabetes: United States. NCHS Data Brief. 2015;2016:1–8. Franco-Paredes C. Social determinants of health and seasonal influenza
vaccination in adults ≥65 years: a systematic review of qualitative and
8. Verger P, Fressard L, Cortaredona S, Lévy-Bruhl D, Loulergue P, Galtier F, et
quantitative data. BMC Public Health. 2013;13:388. https://doi.org/10.1186/
al. Trends in seasonal influenza vaccine coverage of target groups in France,
1471-2458-13-388.
2006 to 2015: impact of recommendations and 2009 influenza a(H1N1)
28. Verger P, Bocquier A, Vergélys C, Ward J, Peretti-Watel P. Flu vaccination
pandemic. Euro Surveill. 2018;23(48):1700801. https://doi.org/10.2807/1560-
among patients with diabetes: motives, perceptions, trust, and risk culture -
7917.ES.2018.23.48.1700801.
a qualitative survey. BMC Public Health. 2018;18:569. https://doi.org/10.1186/
9. Verger P, Cortaredona S, Pulcini C, Casanova L, Peretti-Watel P, Launay O.
s12889-018-5441-6.
Characteristics of patients and physicians correlated with regular influenza
29. Ouellette JA, Wood W. Habit and intention in everyday life: the multiple
vaccination in patients treated for type 2 diabetes: a follow-up study from
processes by which past behavior predicts future behavior. Psychol Bull.
2008 to 2011 in southeastern France. Clin Microbiol Infect. 2015;21:930.e1–9.
1998;124:54–74. https://doi.org/10.1037/0033-2909.124.1.54.
https://doi.org/10.1016/j.cmi.2015.06.017.
30. Marteau TM, Hollands GJ, Fletcher PC. Changing human behavior to
10. Caille-Brillet AL, Raude J, Lapidus N, De Lamballerie X, Carrat F, Setbon M.
prevent disease: the importance of targeting automatic processes. Science.
Trends in influenza vaccination behaviours--results from the CoPanFlu
2012;337:1492–5. https://doi.org/10.1126/science.1226918.
cohort, France, 2006 to 2011. Euro Surveill. 2013;18:20628.
31. Thomas RE, Lorenzetti DL. Interventions to increase influenza vaccination rates
11. Lo-Ciganic W-H, Donohue JM, Jones BL, Perera S, Thorpe JM, Thorpe CT, et
of those 60 years and older in the community. Cochrane Database Syst Rev.
al. Trajectories of diabetes medication adherence and hospitalization risk: a
2018;5:CD005188. https://doi.org/10.1002/14651858.CD005188.pub4.
retrospective cohort study in a large state Medicaid program. J Gen Intern
32. Osterholm MT, Kelley NS, Sommer A, Belongia EA. Efficacy and effectiveness
Med. 2016;31:1052–60. https://doi.org/10.1007/s11606-016-3747-6.
of influenza vaccines: a systematic review and meta-analysis. Lancet Infect
12. Schwandt A, Hermann JM, Rosenbauer J, Boettcher C, Dunstheimer D,
Dis. 2012;12:36–44. https://doi.org/10.1016/S1473-3099(11)70295-X.
Grulich-Henn J, et al. Longitudinal trajectories of metabolic control from
33. Beyer WEP, McElhaney J, Smith DJ, Monto AS, Nguyen-Van-Tam JS,
childhood to young adulthood in type 1 diabetes from a large German/
Osterhaus ADME. Cochrane re-arranged: support for policies to vaccinate
Austrian registry: a group-based modeling approach. Diabetes Care. 2017;40:
elderly people against influenza. Vaccine. 2013;31:6030–3. https://doi.org/10.
309–16. https://doi.org/10.2337/dc16-1625.
1016/j.vaccine.2013.09.063.
13. Tuppin P, de Roquefeuil L, Weill A, Ricordeau P, Merlière Y. French national 34. Yang L, Nan H, Liang J, Chan YH, Chan L, Sum RWM, et al. Influenza
health insurance information system and the permanent beneficiaries
vaccination in older people with diabetes and their household contacts.
sample. Rev Epidemiol Sante Publique. 2010;58:286–90. https://doi.org/10. Vaccine. 2017;35:889–96. https://doi.org/10.1016/j.vaccine.2017.01.004.
1016/j.respe.2010.04.005. 35. Miller WR, Rollnick S. Motivational interviewing: helping people change. 3rd
14. Cortaredona S, Pambrun E, Verdoux H, Verger P. Comparison of pharmacy-based ed. New York, NY: Guilford Press; 2013.
and diagnosis-based comorbidity measures from medical administrative data.
Pharmacoepidemiol Drug Saf. 2017;26:402–11. https://doi.org/10.1002/pds.4146.
15. Jones BL, Nagin DS, Roeder K. A SAS procedure based on mixture models Publisher’s Note
for estimating developmental trajectories. Sociol Methods Res. 2001;29:374– Springer Nature remains neutral with regard to jurisdictional claims in
93. https://doi.org/10.1177/0049124101029003005. published maps and institutional affiliations.
16. Jones BL, Nagin DS. Advances in group-based trajectory modeling and an
SAS procedure for estimating them. Sociol Methods Res. 2007;35:542–71.
https://doi.org/10.1177/0049124106292364.You can also read
